key: cord-286250-bq1u3d4z authors: Phillips, Joanna J.; Chua, MingMing; Seo, Su-hun; Weiss, Susan R. title: Multiple regions of the murine coronavirus spike glycoprotein influence neurovirulence date: 2001 journal: J Neurovirol DOI: 10.1080/135502801753170273 sha: doc_id: 286250 cord_uid: bq1u3d4z The spike (S) glycoprotein of mouse hepatitis virus (MHV) is a major determinant of neurovirulence. Using targeted recombination we previously demonstrated that the S gene of the highly neurovirulent MHV-4 conferred a dramatic increase in neurovirulence to the mildly neurovirulent MHV-A59. To identify the genetic determinants of neurovirulence within the MHV-4 spike, we generated isogenic recombinant viruses containing various MHV-4/MHV-A59 chimeric spike genes, and studied their phenotypes in vivo. The MHV-4/MHV-A59 chimeric spike genes consisted of either reciprocal exchanges between the S1 and S2 spike subunits, or smaller exchanges specifically in the hyper-variable region (HVR) of S1. The chimeric spike gene containing recombinants all exhibited efficient replication in vitro, yet many were severely attenuated for virulence in vivo. Furthermore, these attenuated recombinants exhibited decreased titers of infectious virus in the brain relative to the parental recombinant viruses containing the full-length MHV-4 or MHV-A59 spike genes. This is the first report that compares the neurovirulence and pathogenesis of isogenic viruses with defined alterations in the MHV spike protein. From these studies, it appears that the interactions of multiple regions of the MHV spike, including the HVR, act in concert to allow for efficient infection of and virulence in the murine central nervous system. Murine coronaviruses, or mouse hepatitis viruses (MHV) , are enveloped positive-stranded RNAviruses that can induce a variety of respiratory, gastrointestinal, and neurologic diseases in rodents. As with a number of viral pathogens, the severity and organ tropism of the disease depend, in part, on the viral strain. Infection with the highly neurovirulent MHV type 4 (MHV-4, a strain JHM isolate) results in a potentially fatal acute encephalomyelitis (Dalziel et al, 1986) . Infection with a similar dose of the mildly neurovirulent MHV type A59 (MHV-A59) results in amuch less severe disease with only mild encephalomyelitis and virtually no mortality (Lavi et al, 1986; Lavi et 01, 1988) . Recently, using targeted recombination, we have demonstrated that the spike gene ofMHV-4 is sufficient to confer this highly neurovirulent phenotype to MHV-A59 (Phillips et al, 1999) . The MHV spike or S glycoprotein, expressed on the virion envelope and on the plasma membrane of infected cells, plays a vital role in viral entry, viral spread, and in the immune response to infection (Collins et 01,1982; Buchmeier et al, 1984; Dveksler etal, 1991; Castro and Perlman, 1995; Bergmann etal, 1996) . Depending both on the viral strain and the host cell the spike is posttranslationally cleaved dividing the spike into an amino-terminal, S1, subunit and a carboxy-terminal, S2, subunit (Sturman et 01, 1990) . Sl, thought to form the globular head of the spike, mediates binding to the viral receptors (Collins et 01, 1982; DeGroot et al, 1987; Kubo et 01, 1994) that are members of the CEACAM(previously Bgp (Beauchemin et al, 1999) ) subclass of the JJ f)hillips et ill Figure I Schemutic diagram of the spike gene from the recombinant viruses and their corresponding virulence following intracranial inoculation. LDc,o assays were conducted as descnhcd previously (Hingley at 01, t (94). The recombinant viruses are represented as indicated. 5,\5,R16 and 5"R29 contain the MHV-i\59 spike and the MHV-4 spike, respectively, 51 4R70 and 71 contain the 51 of the 1'v1l-1V-4 spike with the 52 of the MHV-t\59 spike, and 52,R81 and 82 contains the 51 oflvlHV-t\59 and the 52 of MHV-4. 5.;6.HVRI GO and GI represent recombinants that have a 14Z-amin o acid deletion in the !\-lHV·4 hypervariuble region (HVRJ. while S4HV-!\59R1:J1 and R133 contain an exchange of the IvmV-4 I-IVE for the corresponding MHV'f\59 region. 5 ASOHV-4R51 and R52 contain the MHV-t\5!1 spike with the I-1VR of MI-IV-4. The amino acid residues at the exchange sites are indicated by their position in the MHV-4 spike. Recombinant 52.,R81 contained a single coding mutation LGG1Q. and recombinants 5,HV-A59R131 and R133 each contained a single coding mutation 53591 and N374Y, respectively. Grey bars imlicalc the lvlHV-i\59 sequence, and closed bill'S inrlicate thu MHV-4 seqnonr.e. The 52-amino acid deletion in the IvII-lV·A59 I!VR is indicated. To begin to identify specific regions of the MI-I)' spike which playa role in neurovirulence, we generated recombinant viruses containing exchanges between the S1 and S2 subunits of the highly neurovirulent MHV-4 spike and the much less ncurovirulent lVIHV-A59 spike, The recombinants were generated using a modification of the targeted recombination strategy as recently described by Kuo ei 01 (2000) (see Materials and methods). To reduce the risk of secondary site rnutations influencing the viral phenotype, two independently selentcd recombinants (R), as indicated by their numbering, were isolated and characterized for each desired recombinant virus. As shown in Figure 1, the recombinants containing the St of MHV-4 and the S2 of MI-IV-A59 were named S1 4R70 ancl S1 4R71, and the recombinants containing the S1 of lVIHV-i\5~J and the S2 of MHV-4 were called 82 4R81 and S2 4R82. The S1/82 junction was located just 3' to the cleavage site at amino acid 775 (see Materials and Castro and Perlman, 1995; Borgmann et 01, 1996) . Despite the dramatic differences in ncurovirulenco conferred by spike proteins of lVII-IV-4 and lVIHV-A59 they are highly homologous proteins, sharing 95% amino acid identity in 82 and 90% identity in S1. Within the S1 subunit of the loss nourovirulent MHV-A59 spike, however, there is deletion of 52 amino acids in a region termed tho hypervariable region (HVR) (Luytjes ei ol, 1987) . The spike HVR, loosely defined between amino acids 400 and 600, exhibits a great deal of variation in length from strain to strain (Parker et al, 1989; Banner et 01, 1990) . The highly neurovirulenl lVII-IVA has the longest known HVR, but deletions in this region of up to 159 amino acids have been reported (Parker et 01, 1989) . The specific role of the HVR in pathogenesis appears to be complex and multifunctional. Deletions or mutatiuns in this region have been found to alter viral fusion, eliminate neutralizing monoclonal antibody epitopes, and abrogate CD8+ T-cell epitopes (Dalziel or 01, 1986; Fleming et 01, 1986; Wege et ol, 1988; Parker et ol, 1989; Taguchi and Fleming, 1989; Gallagher ei al, 1990; Pewe et 01,1996; Tsai et 01,1999; Wang et al, 1992) . In addition, deletions or mutations in the HVR have been associated with neuroattenuation. In one such study, a 142-arriino acid deletion in the I-IVR was associated with decreased neurovirulence and decreased viral spread in the CNS (Fazakorloy et aI, 1992) . With the recent development of targeted RNA recombination the structural gones of lVII-IV have become accessible, for the first time, to genetic manipulation (Fischer et 01,1997; Leparc-Goffart et 01,1998) . Using this technique we have definitively demonstrated that the S gene is a major determinant ofMHV neurovirulence (Phillips et 01, 1(99) . To further identify genetic determinants of neuroviru lence within the lVIHV spike, we generated a series of recombinants containing MHV-4/MI-IV-i\59 chimeric spike genes, and examinod the pathogenesis of these isogenic viruses that differed exclusively in spike. From these studies, we conclude that multiple regions of the MHV-4 spike, including the hypervariable region, are required for efficient infection and virulence in the mouse CNS. Methods). As controls, recombinants containing the full-length spike gene ofMHV-4, S4R29, or MHV-A59, SA59R16, were isolated and characterized in parallel. To further decrease the impact of secondary site mutations in the recombinant viruses, we sequenced the entire S gene from at least one of each pair of recombinants. S4R29, SA59R16, and S1 4R71 contained no secondary site mutations in spike. S2 4R81 contained a single point mutation in the si subunit that resulted in a codon change L661Q. An independently isolated recombinant virus, S2 4R82, contained no secondary site mutations in S. To examine the efficiency of replication of the chimeric spike gene recombinants we performed growth curves on L2 cell monolayers with S14R70, S2 4R81, S4R29, and SA59R16. The results are shown in Figure 2A . Consistent with previous reports, recombinants containing the MHV-AS9 spike, SA59R, replicated to higher titer in cell culture than recombinants containing the MHV-4 spike, S4R (Phillips et al, 1999) . Both chimeric spike gene recombinants, S14R70 and S2 4R81, replicated in cell culture, and exhibited similar kinetics as the two parental recombinants, SA69R and S4R.The titers of infectious virus, however, were intermediate between SA6yR and S4R. The plaque morphology of the chimeric spike gene recombinants was also intermediate between SA5yR and S4R. At 48 h after infection of L2 cell monolayers, the S1 4 recombinants and the S2 4 recombinants exhibited a medium-plaque phenotype (0.75 mm and 0.5 mm, respectively) as compared to the large-plaque phenotype produced by SA59R (1.0 mm) and the small-plaque phenotype produced by S4R (0.25 mm), Thus, the Sl/S2 chimeric spike gene recombinants replicated efficiently in cell culture, but exhibited a slightly different in vitro phenotype from either of the parental MHV-4 or MHV-A59 spike containing recombinants. o determine if either the Sl or the S2 subunit of the MHV-4 spike was sufficient to confer an increase in neurovirulence, we infected mice intracranially with S14R70, S1 4R71, S2 4R81, S2 4R82, S4R29. and SAs9R16, monitored for lethality, and calculated lDso values. The neurovirulence of each recombinant is shown in Figure 1 . The LDsos for the S4 and SAS9 recombinants were similar to those reported previously (Phillips et ol, 1999) . The virulence of the chimeric S gene recombinants, however, was very different from either of the parental viruses. Both S1 4 and S2 4 recombinants were highly attenuated for virulence. Consistent with this decrease in virulence, mice exhibited clinical signs such as hunched posture, ruffled fur, arid abnormal gait only at high doses of virus [» 1000 PFU) . To determine if the reduced virulence of the chimeric spike gene recombinants was attributable to decreased viral .raplication in the brain, we titered brain homogenates for infectious, virus at various days after intracranial inoculation with 10 PFU of virus (see Materials and Methods). The results are shown in Figure 3 . As we have shown previously, the titers of infectious virus in the brain were similar following inoculation with S4R and SA59R despite the dramatic difference in virulence exhibited by the two recombinant viruses (Phillips et al, 1999) . The two chimeric S gene recombinants, however. exhibited markedly reduced virus titers. S14R70 exhibited low levels of virus while the amount of infectious S2 4R81 virus was at or below the level of detection (200 PFU/g). Thus, despite the high amino acid identity between the MHV-4 and MHV-A59 spike proteins they could not substitute functionally for one another in an in vivo infection. This result suggests that the high neurovirulence conferred by the MHV-4 spike protein requires specific homotypic interactions between regions in 51 and 52, and that disruption of these interactions has a major impact on the pathogenesis of the virus, Within the 51 subunit there is a region, termed the hypervariable region (HVR), which can tolerate extraordinary variation particularly with respect to its length (Banner et al, 1990; Parker et al, 1989) . It has been proposed that the HVR has a vital function in MHV pathogenesis. Numerous variant viruses have been identified in which mutations or deletions in this region have been associated with alterations in neuropathogenesis (Dalziel et al, 1986; Fleming et al, 1986; Wege et ol, 1988; Parker et al, 1989; Taguchi and Fleming, 1989; Gallagher et al, 1990; Wang et al, 1992) . To definitively assess the role of the hypervariable region (HVR) in neurovirulence, we used targeted recombination to generate a series of recombinant viruses with small exchanges in the HVR between MHV-4 and MHV-A59. To determine if the HVR of the highly neurovirulent MHV-4 spike was necessary to confer an increase in neurovirulence, we generated recombinants in which the MHV-4 HVR was deleted or replaced with the MHV-A59 HVR (see Figure 1 ). Deletion of the MHV-4 HVR was based on a previously identified neutralizing monoclonal antibody escape mutant with a deletion in the HVR, and the two independently isolated recombinants containing this 142amino acid deletion were named 5 4AHVR160 and S4AHVR161 (Dalziel et al, 1986; Parker et al, 1989) . Recombinants S4HV-A59R131 and S4HV-A59R133 contained the MHV-4 5 gene with replacement of a l1Z-amino acid region in the HVR with the corresponding 60-amino acid region from the MHV-A59 S gene. In addition, to examine if the MHV-4 HVR was sufficient to confer some degree of increased neurovirulence we replaced theMHV-A59 HVRwith the MHV-4 HVR (see Figure 1 ). Thus, recombinants 5 A59HV-4R51 and SA59HV-4R52 contained the MHV-A59 spike with replacement of a 150-amino acid region in the HVR with the corresponding 202-amino acid region from the MHV-4 spike. Once again, we sequenced the entire 5 gene from at least one recombinant of each pair of recombinants. Recombinants S4AHVR160 and SA59HV-4R51 contained no secondary site mutations. Sequencing of recombinant 8 4HV-AS9R131, containing the MHV-4 spike with the MHV-A59 spike HVR, revealed a single co~ing mutation S3591. Interestingly we were unable to select a 5 4HV-A59R recombinant without a secondary site mutation. S4HV-A59R133, an independently isolated recombinant, contained a single coding mutation N374Y, and two additional recombinants, S4HV-AS9R132 and R134, were partially sequenced and found to contain single coding mutations (L1114F and L660Q, respectively). These two recombinants were not further characterized. The inability to select S4HV-AS9 recombinants without secondary site mutations in S suggests that these mutations were selected for during replication in cell culture. The replication efficiency of these recombinants on L2 cell monolayers is shown in Figure 2B . The recombinants all exhibited similar kinetics of viral replication; however, the extent to which they replicated differed. In general, the growth phenotype of the recombinants segregated with the spike gene. Recombinants with primarily MHV-4 spike sequence, 5 4AHVR and S4HV-A59R, replicated to a similar extent as S4R, and the HVR recombinant with primarily MHV-A59 spike sequence, SA59HV4R, replicated to a similar extent as S A59 R. The plaque morphology of the recombinants was also dependent on the spike gene. 48 hpj. of L2 cell monolayers, SA 59R, and SA59HV-4R exhibited a large-plaque morphology (1.0 rom). The recombinants with primarily the MHV-4 spike exhibited smaller plaque-phenotypes. S4HV-AS9R exhibited an intermediate-plaque size (0.75 mm), and S4AHVRexhibited a delayed plaque" phenotype as plaques were not yet visible at 48 h p.i, The delay in plaque-formation observed with S4AHVR was consistent with previous reports associating deletions in the MHV-4 HVR with delayed fusion and decreased cytopathicity (Gallagher et al, 1990) . In general, recombinants with alterations in the hypervariable region exhibited similar properties in cell culture as the parental spike from which they were derived. To determine the effect of mutations in the HVR on virulence, we inoculated animals intracranially with serial dilutions of the recombinant viruses, observed them for lethality, and then calculated LD 50 doses. The results are shown in Figure 1 . Removal of the MHV-4 HVR, either by deletion, S4AHVR, or by replacement with the MHV-A59 HVR, 8 4HV-AS9R, resulted in recombinant viruses with attenuated virulence. Thus the MHV-4HVR was necessary for the highly neurovirulent phenotype conferred by the MHV-4 spike. In contrast, insertion of the MHV-4 HVRinto the MHV-A59spike, SA59HV-4, did not alter the virulence of the recombinant viruses indicating that the MHV-4 HVR was not sufficient to confer an increase in virulence to the MHV-AS9 spike. In addition, the similar virulence exhibited by SA59R16 and the two recombinants SA59HV-4R51 and R52 indicate that an insertion of 52 amino acids is well tolerated in the MHV-A59 HVR. To determine the effect of alterations in the HVR on the efficiency of viral replication in the brain, we inoculated mice intracranially with the HVR recombinants, and determined the titers of infectious virus in the brain at various times p.i. (Figure 4 ). Consis-. tent with their poor virulence, the MHV-4 8 gene containing recombinants With deletions or exchanges in the HVR'exhibited dramatically reduced virus titers in the brain which were at or below the level of Rao and Gall 0.0001 for both). detection (200 PFU/g). On days 3 and 5 p.i. with S4~HVR or S41-IV-A59R the titers of infectious virus were significantly different from the titers following infection with S4R29 (two-tailed r-test, P > 0.0001). In contrast, replacement of the MHV-A59 HVR with the corresponding region from MHV-4 did not 1'8SU1[ in a dramatic reduction in the amount of infectious virus recovered from the brain. Although SA5~HV-4R exhibited similar average Liters of infectious virus as S,wJR, there was a large degree of animal to animal variation as indicated in Figure 4 . On clay 7 p.i., some :animals exhibited relatively high titers of infectious virus. To demonstrate that these high tilers late in infection were not attributable to the replication of variant viruses with secondary site mutations in the spike, we amplified virus from the brain al 7 days p.i., and sequenced the spike gene to generate a coni sensus sequence. No secondary site mutations were founel in the spike gene. Despite efficient replication and growth in cell culture most of the chimeric spike containing recombinant viruses exhibited attenuated virulence in vivo. Due to the numerous selective pressures present during an in vivo infection it is possible that small alterations jn viral entry, spread, or replication, which Were not readily observable in cell culture, had a profound impact on virulence. Previous studies have found an association between alterations in the abilty of the MHV spike to induce fusion and a1terations III pathogenesis (Frana et ol, 1985; Gallagher et al, 1990; Gallagher et al, 1991; Gallagher ei al, 1992; Multiple regionsof spike influenceneurovirulence JJ Phillips et al not all of the recombinants may be associated with decreased cell-cell fusion in cell culture. With the development of targeted recombination the structural genes of MHV have become amenable to genetic manipulation. Using this technique we previously demonstrated that the spike gene of the highly neurovirulent MHV-4 virus can confer a dramatic increase in virulence to the much less virulent MHV-A59 virus (Phillips et 01, 1999) . Here we report on studies designed to localize this phenotype to a specific region of the spike, and to examine the effect of specific alterations in the spike on pathogenesis. Using targeted recombination, we generated a series of recombinants containing MHV-4/MHV-A59 chimeric spike genes. We first characterized their in vitro phenotype to ensure that the chimeric spike genes were functional for infection and replication in tissue culture. We then examined the in vivo phenotype of the recombinants. The two parental recombinants, 8 4R (containing the MHV-4 spike gene) and 8 AS9R (containing the MHV-A59 spike gene), exhibit similar kinetics of replication in cell culture, but 8 AS9R achieves higher peak titers than 8 4 R. In cell culture, all of the chimeric spike gene recombinants replicated efficiently. They exhibited similar kinetics of replication as 8 4R and 8 AS9R, and their peak titers were similar to or intermediate between the high titers observed with 8 AS9R and the lower titers observed with 8 4 R. When we infected mice with the recombinant viruses, however, we observed that many of the recombinants displayed a different phenotype from either of the two parental recombinants. Recombinants containing exchanges between the 81 and 82 subunits (81 4R or 82 4R) or deletion or replacement of the MHV-4 spike hypervariable region (84~HVR or 8 4HV-A59R, respectively) exhibited a profound decrease in viral replication in the brain as demonstrated by the titers of infectious virus. Furthermore, these recombinants were highly attenuated for neurovirulence as demonstrated by their high intracranial LD so values. The high titers of infectious virus in the brains of mice infected with recombinant 8A59HV-4R, containing the MHV-A59 spike with the MHV-4 spike HVR, demonstrated that not all of the chimeric spike gene recombinants exhibited attenuated replication. In addition, the virulence of 8 AS9HV-4R was not attenuated as it was similar to the parental SA59R rec~mbinant. As illustrated by the two parental recombinants decreased neurovirulence is not always associated with decreased titers of infectious virus in the brain. S4R and SAS9R exhibit a 3 IOg10 difference in intracranial LD so, yet when mice are infected with the same dose of virus, S4Rand 8 AS9R exhibit similar titers of infectious virus in the brain. The chimeric spike gene recombinants, however, demonstrate that alterations in both Sl and S2 can result in reduced viral titers in the brain, and that this decrease in viral replication in the brain is associated with decreased neurovirulence. The neuroattenuated phenotype of recombinants containing 81/S2 exchanges between the MHV-4 and MHV-A59 8 genes suggest that under the stringent conditions of an in vivo infection homotypic 81/S2 interactions are required for efficient replication in the eNS. Moreover, in combination with previous studies associating mutations in si and 82 with neuroattenuation, these data suggest that regions within both subunits of the MHV-4 spike are required to confer the highly neurovirulent phenotype (Parker et al, 1989; Wang et al, 1992; Grosse and Siddell, 1994; 8aeki et 01,1997) . These regions of spike, which when mutated, deleted, or replaced influence neuropathogenesis, may interact structurally in the tertiary conformation of the protein, or cooperate functionally for efficient functioning of spike. There is experimental evidence to support the idea that regions of 81 and S2 closely interact. Variant viruses have been identified in which mutations in S'l, responsible for receptor binding, alter fusion, and mutations in S2, responsible for viral fusion, alter the conformation of Sl or confer resistance to neutralization by soluble receptor (Gallagher et al, 1990; Grosse and Siddell, 1994; Saeki et al, 1997) . Although the spike proteins of MHV-4 and MHV-A59 share a high degree of amino acid homology, the severely attenuated phenotype of the recombinants with chimeric spike genes indicate that there are critical differences between the two spike genes in both Sl and S2. As has been demonstrated with Theiler's virus (Pritchard et al, 1993; Zhang et al, 1993) , the in vivo attenuation observed with viruses containing chimeric viral proteins between similar strains demonstrates the complexity and sequence specificity of viral protein interactions, and attests to the stringent constraints imposed on a virus in vivo. Additional recombinants containing alterations exclusively in the MHV-4 hypervariable region (HVR) demonstrated that this region of Sl is required for the high neurovirulence of the MHV-4 spike. Deletion or replacement of the MHV-4 HVR resulted in neuroattenuation of the recombinant viruses. These results definitively confirm the data from previous studies suggesting an association between mutations or deletions in the MHV-4 HVR and neuroattenuation (Dalziel et al, 1986; Fleming et al, 1986; Wege et al, 1988; Parker et al, 1989; Taguchi and Fleming, 1989; Gallagher et al, 1990; Wang et al, 1992) . Although no single function has been attributed to the HVR, studies of variant viruses suggest that alterations in this region can influence 81-82 interactions and fusion (Gallagher et al, 1990; Gallagher, 1997) . In ad-'dition, the MHV-4 spike HVR contains neutralizing antibody epitopes and T-cell epitopes (Dalziel et aI, 1986; Taguchi and Fleming, 1989; Gallagher et al, 1990; Castro and Perlman, 1995) , and thus alterations in this region could also influence the immune response to infection. Due to the severely attenuated in vivo replication and virulence exhibited by recombinants with deletion or replacement of the MHV-4 spike HVR, however, it appears that this region is critical for the in vivo functioning of the MHV-4 spike. In contrast, replacement of the MHV-A59 spike HVR with the corresponding region from the MHV·4 spike demonstrated that the HVR of MHV-4 was not sufficient to confer an increase in neurovirulence. Interestingly, the presence of the MHV-4 HVR in the MHV-AS9 spike did not.appear to alter the LD so of the resulting recombinant virus. Thus, not only could the MHV-4 HVR grossly substitute for the corresponding region ofMHV-A59, but the MHV-A59 spike HVR was able to tolerate an insertion of S2 amino acids with no obvious effect on neurovirulence. To identify more subtle alterations or defects in the MHV-4/MHV-AS9 chimeric spike proteins we tested a number of their in vitro properties. Previous studies have suggested that the fusogenicity of the MHV spike is an important factor in both viral spread in the host and in virus-induced cytotoxicity (Gallagher et al, 1990; Gallagher et ol, 1992; Rao and Gallagher, 1998) . To determine if the spike proteins of the attenuated recombinants exhibited an altered ability to induce cell-cell fusion we performed cell-cell fusion assays. Four of the recombinants (S14R70, S4~HVR160. S4HV-AS9R131, and SA59HV-4R51) exhibited some degree of delayed fusion. Although not all of the attenuated recombinants exhibited decreased cell-cell fusion, it appeared that a severe reduction in cellcell fusion was correlated with neuroattenuation. Interestingly, the slight decrease in cell-cell fusion observed with the recombinants containing a replacement of the MHV-AS9 HVR with the MHV-4 HVR did not result in a corresponding decrease in virulence. This result suggests that MHV infection and replication in the eNS may require a certain threshold level offusion. Although not all viruses with fusion defects exhibit neuroattenuation ( ST Hingley, unpublished data) , decreased fusogenicity of the MHV spike has previously been found to associate with decreased spread in the CNS (Gallagher et al, 1990; Fazakerley et al, 1992) . Thus for some ofthe recombinant viruses decreased fusion resulting in decreased spread and dissemination, may contribute to the decreased replication observed in the CNS. In additio,n to cell-cell fusion we also examined the temperature sensitivity and the thermostability of the recombinant viruses. The temperature sensitivity of tJ:le recombinants, as determined by the plating effi-c~ency at 39.5°C compared to at 37.0°C, revealed no dIfferences between the highly virulent and highly atteuuated recombinant viruses (data not shown). The thermo stability of the recombinant viruses was also examined by determining the decrease in viral stock titers over time at 40°C, pH 7.3 (data not shown). The neuroattenuated recombinants fell into a range of thermostabilities, however, they were bounded by the highly thermostabile SA59R16 and much less fuermostabile 8 4R29 (data not shown). Thus. there was no clear correlation between thermostability and neurovirulence. Due to the many selective pressures inherent in an in vivo infection. small differences in the kinetics of an infection can dramatically affect pathogenesis. Such differences may explain why many of the recombinants with alterations in the spike displayed efficient replication in cell culture. yet highly inefficient replication in vivo. By creating recombinant viruses with heterotypic Sl/S2 interactions we may have inadvertently altered either structural or functional interactions between the 81 and 82 subunits. These perturbations in spike may not have been apparent in cell culture due to differences between in vitro and in vivo infection such as receptor density on the host cell, the availability of free virus, the virion density required for an efficient infection, and the fusogenicity of the host cell. It is also possible that there were slight alterations in the processing or expression of the chimeric spike proteins that were not apparent in vitro. The lack of a correlation between replication in cell culture and replication in vivo is not unique to the chimeric spike gene recombinants as it has been observed previously for other MHV strains and for other viruses (Gallagher et ol, 1990; Pritchard et al, 1993) . The targeted recombination technique is a powerful tool for studying MHV pathogenesis as defined alterations in spike can be introduced into isogenic viruses. As with all strains of MHV, point mutations can arise during viral replication. To decrease the impact of such secondary site mutations in the recombinant viruses we selected and characterized two independent recombinants for each desired recombinant, and sequenced the entire S gene from at least one recombinant. Interestingly, S4HV-A59R was the only recombinant for which multiple. independent recombinants were found to contain single coding mutations in S after minimal passage in cell culture. The location of one of these compensatory mutations identified in S2 (Ll114F) is striking as numerous studies have identified alterations in this exact residue. The 1114 codon has been identified by Gallagher et al (1991) (L1114R) as one of three amino acids responsible for conferring pH-dependent fusion on MHV-4, by Saeki et aJ (1997) (L1114F) in the spike gene of soluble-receptor resistant mutants, and by Wang et a1 (1992) (Ll114F) in a neuroattenuated monoclonal antibody escape mutant. It is interesting to speculate that this region of S2 plays a prominent role in the conformation of spike and perhaps in the dynamic interaction that occurs between S1 and S2. The identification of compensatory mutations that arise both in vivo and in vitro as a result of various alterations in spike may provide valuable clues as to which regions of spike undergo functional and/or structural interactions. The spike of MHV is a major determinant of neurovirulence. This is the first report using targeted recombination to examine the role of specific regions of the MHV spike in neuropathogenesis. Despite efficient replication in cell culture many of the MHV-4/MHV-A59chimeric S gene recombinants displayed severely attenuated replication in vivo. Thus, the degree of amino acid homology between the MHV-4 and MHV-A59 spike glycoproteins was high enough for the chimeric spike gene recombinants to replicate efficiently in vitro, yet not high enough for efficient replication in the murine eNS. In addition, recombinants with alterations in the spike HVR have confirmed the importance of this region in the highly neurovirulent phenotype conferred by MHV-4 spike. Thus the pronounced neurovirulence conferred by the MHV-4spike appears to depend on the structural and/or functional interaction of multiple regions of spike rather than on discreet independent functional domains within spike. Feline FCWF cells, fMHV, and AIM were obtained from Paul S Masters (Albany, NY). The fMHV virus is a recombinant MHV that contains the ectodomain of the S protein of feline infectious peritonitis virus, the' Alb4 replicase, and the MHV-A59 3'-end (Kuo et al, 2000) . AIM, derived from MHV-A59, is a temperature-sensitive N gene deletion mutant that produces small plaques at the non-permissive temperature (39°C) and is thermolabile (Koetzner et al, 1992) . Viruses were propagated on either murine 17CI-l cells or feline FCWF cells, and plaque assays and purifications were carried out on murine L2 cells. Cells were maintained on plastic tissue culture flasks in Dulbecco's minimal essential medium (MEM) with 10% fetal bovine serum (FBS). Spinner cultures of L2 cells were maintained in [oklik's MEM with 10% FBS at densities of between 2 x 10 5 and 2 x 10 6 cells per ml. The pMH54 plasmid (obtained from Paul S Masters) includes from codon 28 of the HE pseudogene through to the 3' end of the MHV-A59 genome as described .by Kuo et a1 (2000) . The pGEM4Z plasmid containing the MHV-4 8 gene, p8wt, was obtained from Michael J Buchmeier (Lalolla, CAl (Gallagher et al, 1991) . The pGEM-S4'and pGEM-A59 plasmids were derived from pSwt and contained the MHV-4 and the MHV-A59spike genes, respectively (Phillips et al, 1999) . These plasmids also contained silent mutations at codons 12 and 13 of the S gene to create an AvrIl site, and at 12 nt past the S gene stop codon to create a Sbfl site. These restriction sites were used for shuttling the various spike genes into the pMH54 vector. To generate the various exchanges in thelS gene, coding-silent changes were introduced into the MHV·4 and MHV-A59 spike genes in pGEM-S4 or pGEM-A59, respectively, that created unique andlor common restriction sites. DNA manipulations were carried out using standard methods (Maniatis et al, 1982) . All cloning sites and regions generated by peR mutagenesis were verified by sequencing. and the composition of each construct was checked by restriction analysis. To exchange the MHV-4 and MHV-A59 81 and S2 subunits an EcoRV site was introduced at codon 775 (MHV-4 sequence). Avrll/EcoRV or EcoRV/SbfI segments were then exchanged between the MHV-4and MHV-A59 spikes. The resulting plasmids, pG-S1 4 , containing the Sl of MHV-4 and the S2 of MHV-A59, and pG-82 4 , containing the Sl of MHV-A59 and the 82 of MHV-4, were digested with Avrll/SbfI and the chimeric S genes were shuttled into the pMH54 plasmid creating pMH54-814 and pMH54, 52 4, A S gene with the MHV-A59 spike and the MHV-4 HVR was generated by introducing silent mutations into the MHV-4 spike creating a Pstl site at codon 412 and a KpnI site at codon 614. Using these two sites the resulting 606-nt MHV-4 fragment was cloned into the MHV-A59 spike creating plasmid pG-S A59HV-4. The Avrll/SbfI segment from this' plasmid was then cloned into pMH54 creating plas mid pMH54-SA59HV-4. The MHV-A59 HVR was in troduced into the MHV-4 spike by introducing silenm utations in the MHV-A59 spike at codons 488 auG! 548 creating an EcoR47ill site and a Bglll site, respeca tively. For cloning purposes a silent mutation creati! ing a Bglll site was also introduced into the MHV.{4} S gene at codon 600. Using the Ec047ill and BgIN! restriction sites the MHV-A59 HVR was cloned intG: the corresponding region of the MHV-4 spike create ing plasmid pG-S 4HV-A59. The Avrll/Sbfl segment from this plasmid was then cloned into pMH54 creã ting pMH54-S 4HV-A59. Deletion of the MHV-4 H\lB was based on a previously identified deletion in Wy. MHV spike (Dalziel et ol, 1986; Parker et al, 1989 ): The 142-amino acid deletion between codons 43jl: and 575 and codon substitution at K433N was crã ted using PCR mutagenesis in pGEM-S4. The rer' sulting plasmid, pG-S4~HV, was used to shuttle thà vrll/Sbfl region of S into pMH54 creating pMH54a For RNA transcription the designated pMH54 pla~' mids were linearized just 3' to the poly(A) tail by di· gestion with Pacl, Targeted RNA recombination and selection of recombinants , Targeted recombination was carried out, as describe, by Kuo (2000) , between the recipient virus fMHVan· synthetic capped RNAs transcribed from. pMH54-54, pMH54, pMH54-S1 4 , pMH54-S2 4 , pMH54-S 4HV-A59, and pMH54-S 4 .6. . Recombination with RNA transcribed from pMH54-SAs9HV-4 was carried out with the Alb4 recipient virus as described previously (Masters et al, 1994; Leparc-Goffart et aI, 1998; Phillips et al, 1999) . The potential recombinants were plaque-purified and screened, using PCR amplification and gel electrophoresis, for the presence of the 5' portion of the recombinant S gene (Fischer et aI, 1997; Leparc-Goffart et ol, 1998; Phillips et al, 1999; Kuo et al, 2000) . The selected recombinants were then plaque-purified once again, and viral stocks were grown on 17CI-1 cells for further analysis (Fischer et al, 1997) . For each desired recombinant virus at least two recombinants derived from independent recombination events were characterized. S gene sequencing For sequencing of the 8 gene, cytoplasmic RNA extracted from virus-infected L2 cells was used as a template for reverse transcriptase-mediated PCR (RT-PCR) amplification. The oligonucleotides described previously (Leparc-Goffart et 01, 1998; Phillips et al, 1999) were used for amplification of the designated regions. Double-stranded peR products were analyzed by automated sequencing by the Taq dye terminator procedure according to the manufacturer's protocol (Taq DyeDeoxy Terminator Cycle Sequencing kit; Applied Biosystems). The entire S gene was sequenced from at least one member of each pair of recombinant viruses. The sequence of each S gene was identical to the previously published sequences for the MHV-4 and MHV-A59 S genes Parker et 01, 1989; Gallagher et al, 1991; Phillips et al, 1999) except for recombinants containing the 82 ofMHV-4, S2 4R, and recombinants containing the MHV-4 spike with the MHV-A59 hypervariable region, S4HV-A59R. S2 4R81 contained a single point mutation in the 81 subunit that resulted in a codon change L661Q. S2 4R82, an independently isolated recombinant virus, contained no secondary site mutations in S. S4HV-A59R131 contained a single coding mutation S3591. An independently isolated recombinant, S4HV-A59R133, also contained a single coding mutation N374Y. Two additional recombinants, S4HV-A59R132 and R134, were partially sequenced and found to contain single coding mutations (L1114F and L660Q, respectively). Virus growth curves onfluent L2 cell monolayers in 12-well plates werẽ fected with each virus at the indicated multiplic-Ity of infection. Infections were carried out in duplicate. Following adsorption for 1 h at 37°C, the cells Were washed with Tris-buffered saline three times and then fed with 1.5 mL ofDMEM-l0% FBS. At the specified times, the cells were lysed by three cycles of freeze-thawing, and the supernatants were removed and titered by plaque assay on L2 cells as previously described (Gombold et al, 1993) . Animals qJ7BlI6 (B6) mice were purchased from the National Cancer Institute (Frederick, MD). Four-week-old, MHV-free male mice were used in all experiments. Mice were anesthetizedwith isoflurane (IsoFlo; Abbott Laboratories, North Chicago, IL). The amount of virus designated in each experiment was diluted in PBS containing 0.75% bovine serum albumin, and a total volume of 20 ILl was injected into the left cerebral hemisphere. Mock-infected controls were inoculated similarly but with an uninfected 17CI-l cell lysate at a comparable dilution. Fifty percent lethal dose (LDso) ..assays were carried out as described previously (Hingley at al, 1994) . Mice were inoculated intracranially with 4-or 5-fold serial dilutions of recombinant viruses. A total of 5 animals per dilution per virus were analyzed. Mice were examined for signs of disease or death on a daily basis up to 21 days p.i, LD so values were calculated by the Reed-Muench method (Reed and Muench, 1938; Smith and Barthold, 1997) . The levels of infectious virus in the CNS as a function of time following infection with the recombinant viruses was determined in mice inoculated intracranially with 10 PFU of virus. On days 1, 3, 5, and 7 postlnfsction, mice were sacrificed, perfused with 10 ml of PBS, and the brains were removed. The left-half of the brain was placed directly into 2 ml of isotonic saline with 0.167% gelatin (gel saline) (Ramig, 1982) . All organs were weighed and stored frozen at -BO°C until titered for virus. Brains were homogenized, and virus titers were determined by plaque-assay on L2 cell monolayers (Ringley et al, 1994) . The titers of infectious virus in the brain were determined for all of the recombinant viruses at once in three independent experiments. To quantitate cell-cell fusion induced by the recombinant viruses, L2 cell monolayers ,in 6-well plates were infected with virus in duplicate (MOl 0.1 or 0.7 PFU/cell) and incubated for 1 h at 37°C. Following adsorption, the cells were washed with 'Iris-buffered saline twice and then fed with 3 ml of DMEM-l0% FBS. At 7 and 9 h p.i., the cells were washed with PBS and fixed with 2% paraformaldehyde (Sigma Chemical Co, St. Louis, MO). For each 20x-objective field under phase-contrast microscopy the total number of nuclei in syncytia was counted. At least 10 random fields were counted per well, and the mean number of nuclei in syncytia per field was calculated. The percent fusion relative to that observed with S4R29 was calculated for each experiment. To demonstrate that each 20x-objective field had a similar number of cells, the total number of cells per field was counted at random. The results. shown in Figure 5 . represent the means (and standard deviations) of at least three independent experiments except S46HVR160. which is from two independent experiments. A clustering of RNA recombination sites adjacent to a hypervariable region of the peplomer gene of murine coronavirus Redefined nomenclature for members of the carcinoembryonic antigen family The JHM strain of mouse hepatitis virus induces a spike proteinspecific Db.restricted cytotoxic T cell response Murine hepatitis virus-a (strain JHM)-induced neurologic disease is modulated in vivo by monoclonal antibody CD8+ T-cell epitopes within the surface glycoprotein of a neurotropic coronavirus and correlation with pathogenicity Monoclonal antibodies to murine hepatitis virus-a (strain JHM) define the viral glycoprotein responsible for attachment and cell-cell fusion Site-specific alteration of murine hepatitis virus type 4 peplomer glycoprotein E2 results in reduced neurovirulence Evidence for a coiledcoil structure in the spike proteins of coronaviruses Cloning of the mouse hepatitis virus (MHV)receptor: Expression jn human and hampster cell lines confers susceptibility to MHV 1 envelope glycoprotein deletion mutant of mouse hepatitis virus type-4 is neuroattenuated by its reduced rate of spread in the central nervous system Analysis of a recombinant mouse hepatitis virus expressing a foreign gene reveals a novel aspect of coronavirus transcription Antigenic relationships of murtna coronaviruses: Analysis using monoclonal antibodies to JHM (MHV-4) virus Pathogenicity of antigenic variants of murine coronavirus JHM selected with monoclonal antibodies Proteolytic cleavage of the E2 glycoprotein of murine coronavirus: Host-dependent differences in proteolytic cleavage and cell fusion A role for naturally occurring variation of the murine coronavirus spike protein in stabilizing association with the cellular receptor Cell receptor-independent infection by a neurotropic murine coronavirus Alteration of pH dependence of coronavirus-induced cell fusion Effect of mutations in the spike glycoprotein Neutralization-resistant variants of a neurotropic coro navirus are generated by deletions within the aminoter minal half ofthe spike glycoprotein Fusion-defective mutants of mouse hepatitis virus A59 contain a mutation in the spike protein cleavage signal Single amino acid changes in the SZ subunit of the MHV surface glycoprotein confer resistance to neutralization by Sl subunit-specific man oelonal antibody MHV A59 fusion mutants are attenuated and display altered hepatotropism Repair and mutagenesis of the genome of a deletion mutant of the murine coronavirus mouse hap atitis virus by targeted RNA recombination Localization ofneu tralizing epitopes and the receptor-binding site withiJ1 the amino-terminal 330 amino acids of the murine coro: navirus spike protein Retargeting of coronavirus by substitution of the spike glycoprotein ectodomain: Crossing the host cell species barrier Limbic encephalitis after inhalation of a murine coronavirus The organ tropism of mouse hepatitis virus strain A59 is dependent on dose and route of inoculation Targeted recombination within the spike gene of murine coronavirus mouse hepatitis virus-A59: Q159 is a determinant of hepatotropism Primary structure of the glycoprotein E2 of coronavirus MHV-A59 and identification ofthe trypsin cleavage site Molecular cloning, a laboratory manual Optimization of targeted RNA recombination and mapping of a novel nucleocapsid gene mutation in the corona virus mouse hepatitis virus Sequence analysis reveals extensive polymorphism and evidence of deletions within the E2 glycoprotein gene of several strains of murine hepatitis virus Cytotoxic T cell-resistant variants are selected in a virus-induced demyelinating disease Pathogenesis of chimeric MHV4/MHV-A59 recombinant viruses: The murine coronavirus spike protein is a major determinant of neurovirulence Assembly of Theiler's virus recombinants used in mapping determinants ofneurovirulence Isolation and genetic characterization of temperature sensitive mutants of simian rotavirus SAll Intracellular complexes of viral spike and cellular receptor accumulate during cytopathic murine coronavirus infections A simple method of estilflating fifty per cent points Identification of spike protein residues of murine coronavirus responsible for receptor-binding activity by use of soluble receptor-resisitant mutants Methods in viral pathogenesis Conformational change of the coronavirus peplomer glycoprotein at pH 8.0 and 37°C correlates with virus aggregation and virusinduced cell fusion Comparison of six different murine coronavirus JHM variants by monoclonal antibodies against the E2 glycoprotein A 12-amino acid stretch in the hypervariable region of the spike protein 81 subunit is critical for cell fusion activity of mouse hepatitis virus Sequence analysis of the spike protein gene of murine coronavirus variants: Study ofgenetic sites affecting neuropathogenicity The peplomer protein E2 of coronavirus JHM as a determinant of neurovirulence: Definition of critical epitopes by variant analysis Monoclonal antibodies to the peplomer glycoprotein of coronavirus mouse hepatitis virus identify two subunits and detect a conformational change in the subunit released under mild alkaline conditions Receptor for mouse hepatitis virus is a member of the carcinoembryonic antigen family of glycoproteins Chimeric cDNA studies of Theiler's murine encephalomyelitis virus neurovirulence This work was supported by Public Health Service grants NS-30606 and NS-21954. J.J.P. was supported in part by training grant GM-072Z9.We thank Paul S Masters, Lili Kuo, and Peter 1M 'Rottier for pMH54, Alb4, and fMHV. We thank Jean Tsai for providing us with SA5sR16, Danielle Linn Letting for her help, and Paul Bates. Amy Matthews, and Jean Tsai for critical reading of the manuscript.